hw4

Data section

The data available is from the smart meters and plug-level devices of three households, covering power consumption over a period of time. The primary variable of interest in the smart meter data is the sum of real power over all power phases consumed in the household, while the plug data provides appliance-level consumption. I will be focusing on the smart meter data of household 4 specialized in kitchen (measure type 01: Fridge, 02: Kitchen appliances, 08: Microwave ) from June to December 2012. Specifically, and I am interested in exploring the overall electricity consumption pattern of the household and how it changes over time, with a focus on identifying any trends or seasonality in the data.

Data-science questions

How does the total real power of each household compare over time, and what are the trends?

How does the total electricity consumption of different appliances especially in kitchen for Fridge, Kitchen appliances, and Microwave vary over time in households 4?

Data preparation and EDA

set up dataframe of electricity consumption of different appliances especially in kitchen for Fridge, Kitchen appliances, and Microwave vary over time in households

Code

import pandas as pd
import numpy as np
import glob
import os
import plotly.graph_objects as go
import altair as alt
import plotly.io as pio

pio.renderers.default = "plotly_mimetype+notebook_connected"

Code

plug01_files = glob.glob("./eco/04/01/" + "*.csv")
plug02_files = glob.glob("./eco/04/02/" + "*.csv")
plug08_files = glob.glob("./eco/04/08/" + "*.csv")

plug01 = []
# loop over the list of csv files read in all plugs data for each appliances
for f in plug01_files:
    # read the csv file
    df01 = pd.read_csv(f,names=["measurement"])
# replace all missing data with 0
    df01 = df01.replace(-1, 0)
# calculate the sum of all day measurement
    total01 = df01["measurement"].sum()
# combine measurement sums together 
    plug01.append(total01)

df01 = pd.DataFrame(plug01)
df01.columns = ["Fridge measurement"]
# create a new column with dates starting from 06/27/2012

df01['date'] = pd.date_range(start='2012-06-27', periods=len(df01), freq='D')


# create a list of missing dates
missing_dates = ['2012-09-06', '2012-09-07', '2012-09-08', '2012-09-09', '2012-09-10', '2012-10-26', '2012-10-27', '2012-10-28', '2012-10-29', '2012-10-30', '2012-10-31', '2012-11-01', '2012-11-02', '2012-11-03', '2012-11-04', '2012-11-05', '2012-11-06']

# create a new dataframe with a complete range of dates excluding the missing dates
date_range = pd.date_range(start='2012-06-27', end='2013-01-23', freq='D')
complete_dates = pd.Index(date_range)
missing_dates_index = pd.Index(missing_dates)
valid_dates = complete_dates.difference(missing_dates_index)
df_date_range = pd.DataFrame({'date': valid_dates})
# drop the original date column in df01
df01.drop('date', axis=1, inplace=True)

# set the date column as the index of df_date_range
df_date_range.set_index('date', inplace=True)
# add the date column in df_date_range to df01
df01['date'] = df_date_range.index

# reorder the columns in df01 with the date column at the first position
df01 = df01.reindex(columns=['date'] + list(df01.columns[:-1]))

Code

plug02 = []
# loop over the list of csv files read in all plugs data for each appliances
for f in plug02_files:
    # read the csv file
    df02 = pd.read_csv(f,names=["measurement"])
# replace all missing data with 0
    df02 = df02.replace(-1, 0)
# calculate the sum of all day measurement
    total02 = df02["measurement"].sum()
# combine measurement sums together 
    plug02.append(total02)

df02 = pd.DataFrame(plug02)
df02.columns = ["Kitchen appliances measurement"]
# df02

	Kitchen appliances measurement
0	6.883157e+05
1	1.695786e+06
2	8.464724e+05
3	7.781294e+05
4	7.006198e+05
...	...
189	9.080765e+05
190	5.646038e+05
191	2.163583e+06
192	7.764368e+05
193	5.080496e+05

194 rows × 1 columns

Code

plug08 = []
# loop over the list of csv files read in all plugs data for each appliances
for f in plug08_files:
    # read the csv file
    df08 = pd.read_csv(f,names=["measurement"])
# replace all missing data with 0
    df08 = df08.replace(-1, 0)
# calculate the sum of all day measurement
    total08 = df08["measurement"].sum()
# combine measurement sums together 
    plug08.append(total08)

df08 = pd.DataFrame(plug08)
df08.columns = ["Microwave measurement"]
# df08

	Microwave measurement
0	1.517307e+06
1	1.471807e+06
2	5.028336e+06
3	1.406613e+06
4	4.126534e+05
...	...
189	4.988689e+05
190	5.094321e+05
191	1.108172e+06
192	1.008480e+06
193	7.885643e+05

194 rows × 1 columns

convert Joules per day to kWh per day for all the measurement column

Code

# concatenate columns from df01, df02, and df08 together
df_combined = pd.concat([df01, df02, df08], axis=1)

# convert Joules per day to kWh per day for the Fridge measurement column
df_combined['Fridge measurement'] = df_combined['Fridge measurement'] / 3600000

# convert Joules per day to kWh per day for the Kitchen appliances measurement column
df_combined['Kitchen appliances measurement'] = df_combined['Kitchen appliances measurement'] / 3600000

# convert Joules per day to kWh per day for the Microwave measurement column
df_combined['Microwave measurement'] = df_combined['Microwave measurement'] / 3600000

# round all measurements to 2 decimal places
df_combined = df_combined.round(2)

# print the resulting dataframe
print(df_combined)

# save df_combined as a CSV file
df_combined.to_csv("combined_data.csv", index=False)

          date  Fridge measurement  Kitchen appliances measurement  \
0   2012-06-27                0.73                            0.19   
1   2012-06-28                0.65                            0.47   
2   2012-06-29                0.61                            0.24   
3   2012-06-30                0.86                            0.22   
4   2012-07-01                0.77                            0.19   
..         ...                 ...                             ...   
189 2013-01-19                0.54                            0.25   
190 2013-01-20                0.52                            0.16   
191 2013-01-21                0.52                            0.60   
192 2013-01-22                0.51                            0.22   
193 2013-01-23                0.38                            0.14   

     Microwave measurement  
0                     0.42  
1                     0.41  
2                     1.40  
3                     0.39  
4                     0.11  
..                     ...  
189                   0.14  
190                   0.14  
191                   0.31  
192                   0.28  
193                   0.22  

[194 rows x 4 columns]

Code

plug04_files = glob.glob("./eco/04_sm_csv/04/" + "*.csv")
plug05_files = glob.glob("./eco/05/" + "*.csv")
plug06_files = glob.glob("./eco/06/" + "*.csv")

Since we are gonna compare all household phase real power over time, we should read in all sums of powerallphases

Code

plug04 = []
# loop over the list of csv files read in all powerallphases real power
for f in plug04_files:
    # read the csv file
    df04 = pd.read_csv(f,usecols=[0],names=["real power"])
# replace all missing data with 0
    df04 = df04.replace(-1, 0)
# calculate the sum of powerallphases real power
    total04 = df04["real power"].sum()
# combine real power sums together 
    plug04.append(total04)


df04 = pd.DataFrame(plug04)
df04.columns = ["real power"]
df04['label'] = 'household4'


df04['date'] = pd.date_range(start='2012-06-27', periods=len(df04), freq='D')
df04 = df04.reindex(columns=['date'] + list(df04.columns[:-1]))
df04

	date	real power	label
0	2012-06-27	6.937277e+07	household4
1	2012-06-28	5.726639e+07	household4
2	2012-06-29	5.862045e+07	household4
3	2012-06-30	5.672369e+07	household4
4	2012-07-01	6.414139e+07	household4
...	...	...	...
214	2013-01-27	5.124073e+07	household4
215	2013-01-28	4.517653e+07	household4
216	2013-01-29	2.806189e+07	household4
217	2013-01-30	3.844564e+07	household4
218	2013-01-31	3.685977e+07	household4

219 rows × 3 columns

Code

plug05 = []
# loop over the list of csv files read in all powerallphases real power
for f in plug05_files:
    # read the csv file
    df05 = pd.read_csv(f,usecols=[0],names=["real power"])
# replace all missing data with 0
    df05 = df05.replace(-1, 0)
# calculate the sum of powerallphases real power
    total05 = df05["real power"].sum()
# combine real power sums together 
    plug05.append(total05)


df05 = pd.DataFrame(plug05)
df05.columns = ["real power"]
df05['label'] = 'household5'
df05['date'] = pd.date_range(start='2012-06-27', periods=len(df05), freq='D')

# create a list of missing dates
missing_dates = ['2012-09-07', '2012-09-08', '2012-09-09', '2012-09-10']

# create a new dataframe with a complete range of dates excluding the missing dates
date_range = pd.date_range(start='2012-06-27', end='2013-01-31', freq='D')
complete_dates = pd.Index(date_range)
missing_dates_index = pd.Index(missing_dates)
valid_dates = complete_dates.difference(missing_dates_index)
df_date_range = pd.DataFrame({'date': valid_dates})
# drop the original date column in df01
df05.drop('date', axis=1, inplace=True)

# set the date column as the index of df_date_range
df_date_range.set_index('date', inplace=True)
# add the date column in df_date_range to df01
df05['date'] = df_date_range.index

# reorder the columns in df01 with the date column at the first position
df05 = df05.reindex(columns=['date'] + list(df05.columns[:-1]))
df05

	date	real power	label
0	2012-06-27	6.830255e+07	household5
1	2012-06-28	5.234815e+07	household5
2	2012-06-29	5.594724e+07	household5
3	2012-06-30	5.358165e+07	household5
4	2012-07-01	4.895202e+07	household5
...	...	...	...
210	2013-01-27	6.760786e+07	household5
211	2013-01-28	6.607035e+07	household5
212	2013-01-29	6.384256e+07	household5
213	2013-01-30	6.089343e+07	household5
214	2013-01-31	9.250447e+07	household5

215 rows × 3 columns

Code

plug06 = []
# loop over the list of csv files read in all powerallphases real power
for f in plug06_files:
    # read the csv file
    df06 = pd.read_csv(f,usecols=[0],names=["real power"])
# replace all missing data with 0
    df06 = df06.replace(-1, 0)
# calculate the sum of powerallphases real power
    total06 = df06["real power"].sum()
# combine real power sums together 
    plug06.append(total06)


df06 = pd.DataFrame(plug06)
df06.columns = ["real power"]
df06['label'] = 'household6'


df06 = pd.DataFrame(plug06)
df06.columns = ["real power"]
df06['label'] = 'household6'
df06['date'] = pd.date_range(start='2012-06-27', periods=len(df06), freq='D')

# create a list of missing dates
import pandas as pd

start_date = "2012-11-12"
end_date = "2013-01-03"

date_range = pd.date_range(start=start_date, end=end_date)
missing_dates = ['2012-11-12','2012-11-13','2012-11-14','2012-11-15','2012-11-16','2012-11-17','2012-11-18','2012-11-19','2012-11-20','2012-11-21','2012-11-22','2012-11-23','2012-11-24','2012-11-25','2012-11-26','2012-11-27','2012-11-28','2012-11-29','2012-11-30','2012-12-01','2012-12-02','2012-12-03','2012-12-04','2012-12-05','2012-12-06',
'2012-12-07','2012-12-08','2012-12-09','2012-12-10','2012-12-11','2012-12-12','2012-12-13','2012-12-14','2012-12-15','2012-12-16','2012-12-17','2012-12-18','2012-12-19','2012-12-20','2012-12-21','2012-12-22','2012-12-23','2012-12-24','2012-12-25','2012-12-26','2012-12-27','2012-12-28','2012-12-29',
'2012-12-30','2012-12-31','2013-01-01','2013-01-02','2013-01-03']
# create a new dataframe with a complete range of dates excluding the missing dates
date_range = pd.date_range(start='2012-06-27', end='2013-01-31', freq='D')
complete_dates = pd.Index(date_range)
missing_dates_index = pd.Index(missing_dates)
valid_dates = complete_dates.difference(missing_dates_index)
df_date_range = pd.DataFrame({'date': valid_dates})
# drop the original date column in df01
df06.drop('date', axis=1, inplace=True)

# set the date column as the index of df_date_range
df_date_range.set_index('date', inplace=True)
# add the date column in df_date_range to df01
df06['date'] = df_date_range.index

# reorder the columns in df01 with the date column at the first position
df06 = df06.reindex(columns=['date'] + list(df06.columns[:-1]))
df06

	date	real power	label
0	2012-06-27	1.358194e+07	household6
1	2012-06-28	1.225251e+07	household6
2	2012-06-29	1.279116e+07	household6
3	2012-06-30	2.590369e+07	household6
4	2012-07-01	1.183171e+07	household6
...	...	...	...
161	2013-01-27	1.544921e+07	household6
162	2013-01-28	1.051515e+07	household6
163	2013-01-29	9.630319e+06	household6
164	2013-01-30	1.209106e+07	household6
165	2013-01-31	9.503460e+07	household6

166 rows × 3 columns

convert Joules per day to kWh per day for all the real power column

Code

combined_df = pd.concat([df04, df05, df06], ignore_index=True)
combined_df['real power'] = combined_df['real power'] / 3600000

# round all measurements to 2 decimal places
combined_df = combined_df.round(2)


# save df_combined as a CSV file
combined_df.to_csv("combined_df.csv", index=False)
combined_df

	date	real power	label
0	2012-06-27	19.27	household4
1	2012-06-28	15.91	household4
2	2012-06-29	16.28	household4
3	2012-06-30	15.76	household4
4	2012-07-01	17.82	household4
...	...	...	...
595	2013-01-27	4.29	household6
596	2013-01-28	2.92	household6
597	2013-01-29	2.68	household6
598	2013-01-30	3.36	household6
599	2013-01-31	26.40	household6

600 rows × 3 columns

Results section

A rationale for your design decisions

The first visualization could show the electricity consumption over time for all kitchen appliances measured with smart plugs in the three households. The x-axis would represent time (e.g. days), and the y-axis would represent power consumption in watts. The different lines on the plot would correspond to the different appliances, and the plot could potentially have a dropdown or interactive legend to toggle the different appliances on and off. This would allow users to explore how the electricity consumption of different kitchen appliances varies over time in the 4 household.

The second visualization could show the aggregate real power (powerallphases) over time for each of the three households. The x-axis would represent time (e.g. in hours or days), and the y-axis would represent the total real power all phases in watts or kWh. The plot could have three lines, one for each household, with different colors or styles to distinguish them. Additionally, the plot could include a rolling average or trendline to highlight any patterns or trends in the data. This visualization would allow users to compare the overall real power of the three households and explore any differences or similarities over time.

Visualization 1

In this code, we first import the necessary modules from Plotly. We then create a subplot for each appliance using the make_subplots function, with one row and three columns. We also set the subplot titles to be ‘Fridge’, ‘Kitchen Appliances’, and ‘Microwave’.

We then plot the data for each appliance using the add_trace method of the go.Scatter class. We specify the x and y data for each trace, as well as a name for each trace.

Next, we set the x-axis title using the update_xaxes method, and set the y-axis title for each subplot using the update_yaxes method. We also add a title to the plot using the update_layout method.

Code

import plotly.graph_objects as go
from plotly.subplots import make_subplots

# create a subplot for each appliance
fig = make_subplots(rows=1, cols=3, subplot_titles=('Fridge', 'Kitchen Appliances', 'Microwave'))

# plot the data for each appliance
fig.add_trace(go.Scatter(x=df_combined['date'], y=df_combined['Fridge measurement'], name='Fridge'), row=1, col=1)
fig.add_trace(go.Scatter(x=df_combined['date'], y=df_combined['Kitchen appliances measurement'], name='Kitchen Appliances'), row=1, col=2)
fig.add_trace(go.Scatter(x=df_combined['date'], y=df_combined['Microwave measurement'], name='Microwave'), row=1, col=3)

# set the x-axis title
fig.update_xaxes(title_text='Date')

# set the y-axis title for each subplot
fig.update_yaxes(title_text='Energy Consumption (kWh/day)', row=1, col=1)
fig.update_yaxes(title_text='Energy Consumption (kWh/day)', row=1, col=2)
fig.update_yaxes(title_text='Energy Consumption (kWh/day)', row=1, col=3)

# add a title to the plot
fig.update_layout(title_text='Energy Consumption by Appliance')

# show the plot
fig.show(renderer='notebook')

This graph creates a time series graph with three traces, one for each energy consumption measurement (fridge, kitchen appliances, and microwave). The x-axis represents the date, and the y-axis represents the energy consumption in kilowatt-hours per day. Each trace has a different color to distinguish the measurements.

Code

import plotly.graph_objects as go

# INITIALIZE GRAPH OBJECT
fig = go.Figure()

# TRACE-1: Fridge measurement
fig.add_trace(
    go.Scatter(
        x=df_combined["date"],
        y=df_combined["Fridge measurement"],
        mode="lines+markers",
        marker=dict(
            color=df_combined["Fridge measurement"],
            size=5,
            symbol="circle",
            line=dict(color="DarkBlue", width=1),
            colorbar=dict(title="Fridge Energy Consumption (kWh per day)")
        ),
        line=dict(color="blue", width=1.5),
        name="Fridge",
        visible=True,
    )
)

# TRACE-2: Kitchen appliances measurement
fig.add_trace(
    go.Scatter(
        x=df_combined["date"],
        y=df_combined["Kitchen appliances measurement"],
        mode="lines+markers",
        marker=dict(
            color=df_combined["Kitchen appliances measurement"],
            size=5,
            symbol="square",
            line=dict(color="DarkGreen", width=1),
            colorbar=dict(title="Kitchen Appliances Energy Consumption (kWh per day)")
        ),
        line=dict(color="green", width=1.5, dash="dot"),
        name="Kitchen appliances",
        visible=False,
    )
)

# TRACE-3: Microwave measurement
fig.add_trace(
    go.Scatter(
        x=df_combined["date"],
        y=df_combined["Microwave measurement"],
        mode="lines+markers",
        marker=dict(
            color=df_combined["Microwave measurement"],
            size=5,
            symbol="diamond",
            line=dict(color="DarkRed", width=1),
            colorbar=dict(title="Microwave Energy Consumption (kWh per day)")
        ),
        line=dict(color="red", width=1.5, dash="dash"),
        name="Microwave",
        visible=False,
    )
)

# SET THEME, AXIS LABELS
fig.update_layout(
    template="plotly_white",
    xaxis_title="Date",
    yaxis_title="Energy Consumption (kWh per day)",
    title="Daily Energy Consumption",
)

# DROPDOWN MENUS
fig.update_layout(
    updatemenus=[
        dict(
            buttons=[
                dict(
                    label="Fridge",
                    method="update",
                    args=[{"visible": [True, False, False]},
                        {"title": "Fridge measurement (days)"}]
                    ),
                dict(
                    label="Kitchen appliances",
                    method="update",
                    args=[{"visible": [False, True, False]},
                        {"title": "Kitchen appliance measurement (days)"}]
                    ),
                dict(
                    label="Microwave",
                    method="update",
                    args=[{"visible": [False, False, True]},
                        {"title": "Fridge measurement (days)"}]
                    ),
            ],
            direction="down",
            showactive=True,
            pad={"r": 10, "t": 10},
            x=0,
            y=1.15,
            xanchor="left",
            yanchor="top",
        )
    ]
)

fig.show(renderer='notebook')

Code

import plotly.graph_objects as go

# INITIALIZE GRAPH OBJECT
fig = go.Figure()

# TRACE-1: Fridge measurement
fig.add_trace(
    go.Scatter(
        x=df_combined["date"],
        y=df_combined["Fridge measurement"],
        mode="lines+markers",
        marker=dict(
            size=5,
            symbol="circle",
            line=dict(color="DarkBlue", width=1),
        ),
        line=dict(color="blue", width=1.5),
        name="Fridge",
        visible=True,
    )
)

# TRACE-2: Kitchen appliances measurement
fig.add_trace(
    go.Scatter(
        x=df_combined["date"],
        y=df_combined["Kitchen appliances measurement"],
        mode="lines+markers",
        marker=dict(
            size=5,
            symbol="square",
            line=dict(color="DarkGreen", width=1),
        ),
        line=dict(color="green", width=1.5, dash="dot"),
        name="Kitchen appliances",
        visible=True,
    )
)

# TRACE-3: Microwave measurement
fig.add_trace(
    go.Scatter(
        x=df_combined["date"],
        y=df_combined["Microwave measurement"],
        mode="lines+markers",
        marker=dict(
            size=5,
            symbol="diamond",
            line=dict(color="DarkRed", width=1),
        ),
        line=dict(color="red", width=1.5, dash="dash"),
        name="Microwave",
        visible=True,
    )
)

# SET THEME, AXIS LABELS
fig.update_layout(
    template="plotly_white",
    xaxis_title="Date",
    yaxis_title="Energy Consumption (kWh per day)",
    title="Daily Energy Consumption",
)


fig.show(renderer='notebook')

Fridge: The electricity consumption of a fridge varies over time in households depending on its age, size, and energy efficiency. In general, older fridges and larger fridges consume more electricity than newer and smaller fridges. The data provided shows that the average electricity consumption of a fridge in a household is around 1-2 kWh per day.

Kitchen appliances: the coffee machine, bread baking machine, and toaster were connected to the same electrical outlet or circuit, then they would be consuming electricity from the same source. This means that the total electricity consumption of these appliances combined would be higher than if they were used separately, as there may be some energy loss due to inefficiencies in the circuit or outlet. It is also important to note that the specific electricity consumption of each appliance may vary depending on factors such as usage time and energy efficiency.

Microwave: The electricity consumption of a microwave oven is relatively low compared to other kitchen appliances. The data provided shows that the average electricity consumption of a microwave in a household is around 0.1-0.2 kWh per day.

In conclusion, the electricity consumption of different appliances, especially in the kitchen, varies over time in households depending on various factors such as age, size, and energy efficiency. It is important for households to consider these factors when purchasing new appliances and to use them efficiently to reduce their overall energy consumption.

Visualization 2

The heatmap of real power consumption is an informative and visually appealing way to present data on household energy usage. The chart uses Altair for visualization, to create a color-coded grid that represents the mean real power consumption for each day of the week in a given month. The data is grouped by household, with each represented in its own column, making it easy to compare energy usage patterns between households. The chart also includes tooltips that provide additional information on each data point, and a title that summarizes the purpose of the visualization. Overall, this chart is a powerful tool for understanding real power consumption trends across multiple households.

Code

import altair as alt
import pandas as pd

# Assuming your data is in a pandas DataFrame named df

# Replace the column names with the actual column names in your DataFrame
date_col = "date"
real_power_col = "real power"
label_col = "label"

# Create a new column for the day of the week
combined_df["day_of_week"] = pd.to_datetime(combined_df[date_col]).dt.day_name()

# Creating the chart
chart = alt.Chart(combined_df).mark_rect().encode(
    x=alt.X(f"month({date_col}):O", title="Month"),
    y=alt.Y("day_of_week:O", title="Day of Week"),
    color=alt.Color(f"mean({real_power_col}):Q", scale=alt.Scale(scheme="viridis"), legend=alt.Legend(title="Mean Real Power")),
    tooltip=[
        alt.Tooltip(f"month({date_col}):O", title="Month"),
        alt.Tooltip("day_of_week:O", title="Day of Week"),
        alt.Tooltip(f"mean({real_power_col}):Q", title="Mean Real Power", format=".2f"),
    ],
    facet=alt.Facet(f"{label_col}:N", title="Household", columns=2),
).properties(
    title="Heatmap of Real Power Consumption",
    width=250,
    height=200,
).configure_axis(
    domainWidth=0.8,
    labelFontSize=12,
).configure_title(
    fontSize=16,
)

chart.display()

The graph below visualizes the real power consumption by household over time. The graph plots the real power consumption by label, with each line representing a different household. The X-axis represents the date, and the Y-axis represents the real power consumption in kW. The color of each line represents the label or the household, and hovering over the graph provides a tooltip with additional information. This graph provides a clear and concise way to compare the real power consumption of different households over time. This will create a line chart with dates on the x-axis and real power on the y-axis, and each line representing the daily real power of a specific household. The tooltip argument adds a tooltip that shows the date, household name, and real power when you hover over each point on the chart. The .interactive() method makes the chart zoomable and pannable. Here are things we did to enhance interactivity: Defined default font size, color, and font for consistency, Set custom titles and subtitles for the chart, Changed the color scheme of the labels to ‘dark2’, Increased the font size of the chart title and axis labels, Configured the chart title to be left-aligned, Set the width and height of the chart to 800x400 pixels.

Code

import altair as alt

# Set default font size and color
axisColor = '#000000'
titleColor = '#000000'
font = "Helvetica Neue"

# Define chart
chart = alt.Chart(combined_df).mark_line().encode(
    x=alt.X('date:T', title='Date', axis=alt.Axis(labelColor=axisColor, titleColor=titleColor, labelFont=font, titleFont=font)),
    y=alt.Y('real power:Q', title='Real Power (kW)', axis=alt.Axis(labelColor=axisColor, titleColor=titleColor, labelFont=font, titleFont=font)),
    color=alt.Color('label:N', title='Label', scale=alt.Scale(scheme='dark2'), legend=alt.Legend(titleFont=font, labelFont=font, titleColor=titleColor, labelColor=axisColor)),
    tooltip=['date:T', 'label:N', 'real power:Q']
).properties(
    title={
        "text": ["Real Power Consumption by Household"],
        "subtitle": ["Plotting real power consumption by label over time"],
        "fontSize": 20,
        "color": titleColor,
        "font": font
    },
    width=800,
    height=400
).configure_title(
    fontSize=24,
    anchor='start'
).configure_axis(
    labelFontSize=14,
    titleFontSize=16
)

# Show chart
chart.interactive()

Summary

From the graphs above, we can see that there is a clear seasonal pattern in the household 4. The power usage appears to be higher in the summer months (June through August) and lower in the winter months (December through February). This could be due to factors such as air conditioning usage in the summer and heating usage in the winter.

Additionally, we can see that there is a general increasing trend in the data over time. This suggests that the household is using more power overall as time goes on, potentially due to an increase in the number of people living in the household or an increase in the number of appliances and electronics in the household.

It’s worth noting that there are some outliers in the household 4, particularly in late December 2012 and early January 2013. These may be due to unusual circumstances such as holiday decorations or extreme weather conditions. Overall, the data for household 4 shows a clear seasonal pattern and a general increasing trend over time.

Looking at the plot, we can see that there is a clear seasonality in the household 5, with consumption increasing during the summer months and decreasing during the winter months. The trend shows a clear upward trend from July to November, followed by a downward trend from November to January. The seasonal component shows a repeating pattern with a period of 7 days, indicating weekly seasonality.

From the plot, we can observe that there is a clear daily seasonality in household 6, with higher energy consumption during the day and lower consumption during the night. We can also see that the seasonality appears to be more pronounced in the earlier part of the data, with a gradual decrease in the amplitude of the seasonality over time.

In addition to the daily seasonality, we can also observe that there are some spikes in energy consumption that occur irregularly throughout the household 6. Some of these spikes are quite large, particularly the one that occurs on 1/6/2013.

Regarding trend, we can see that there are some periods of higher energy consumption (e.g. from late July to early September 2012, and from mid-October to early November 2012), but overall the trend is relatively flat.

Overall, it seems that household 6 has a strong daily seasonality in its energy consumption, with some occasional spikes in energy usage. The trend is relatively flat, with some periods of higher energy consumption but no clear overall upward or downward trend. Household 4 has the highest average real power consumption among the three households, with an average of 19.85 watts. This suggests that this household may have more electrical appliances or devices that consume more power than the other households.

Household 6 has the lowest average real power consumption among the three households, with an average of 4.97 watts. This could indicate that this household has fewer or less power-hungry appliances or devices.

It’s worth noting that the average real power consumption for Household 5 is quite close to that of Household 4, with an average of 18.95 watts. This suggests that these two households may have similar electrical setups or lifestyles.

References